@smooth has addressed this point elsewhere (I think on the previous post I linked in this post). He expressed the opinion that lowering the value of witness payments beyond the current level would endanger the integrity of the blockchain.
I haven't done an analysis of costs and risks myself, but in the absence of such an analysis, I'm willing to defer to his judgement on the matter.
Because one person has an opinion on it, we automatically need to regard it as gospel? And by the way he is a well paid witness, of course he is going to say that! Pulling from multiple places makes sense on a lot of levels as opposed to taking it all from the number one onboarding feature steem has currently, author rewards.
It wasn't so much my opinion as my observation/recollection that the existing budget for witnesses (cut by 80% from the previous budget) was designed by Steemit to achieve a reasonable funding for witnesses even during extreme bear market conditions. It is also on the low end of security budgets for comparable blockchains at 0.85% currently, and already scheduled/programmed to decline to 0.095%. I find the latter to be completely reckless and I believe the decline should be halted within a few years at the most, but given the already-low budget and the preprogrammed decline I don't see a good justification for a another cut now.
Never mind what I'm paid, it is completely immaterial to my Steem holdings alone much less net worth. Top 20 witnesses (not all of whom are as well off as I am) are currently paid, before expenses, about as much as a fast food job. If you want well-qualified people with any real technical skills, and/or the knowledge and experience to make good decisions, that's not "well paid" by any means. It's more like at least 5-10x below what such a skill set commands on the open market. If you want a meaningful security margin to support the model that witnesses "will/won't do ____, because otherwise they would get voted out" then witnesses need to be somewhat overpaid (otherwise who cares about getting voted out?).
I'd rather see that assertion tested against other budget priorities, including other, potentially more effective, forms of marketing. If author rewards were the main thing that Steem needs, it would be a lot more successful by now.
And while we are on the subject, if we lowered witness votes from 30 down to 3 that would also help protect the integrity of the blockchain...